Before you optimise, make sure you’re counting the right thing

A campaign can appear to be improving while the business experiences little change. The dashboard reports more conversions, the cost per conversion falls, and everyone agrees the trend looks healthy. Then someone asks how many of those conversions became genuine enquiries or completed orders. The room gets quieter.

The first strategic job is to establish what the account is actually counting. A form submission, a click on a phone number and a completed sale are different events. Combining them into one headline number can hide the distinction between activity and a useful outcome.

Build an outcome map

Write the customer journey in plain language. For a service business it might be enquiry, valid enquiry, qualified opportunity and acquired customer. For a shop it might be checkout, paid order and fulfilled order after cancellations.

Then map every tracked action to a stage. Record where the event originates, what triggers it, whether it can repeat and which campaigns use it. An event with no clear business interpretation deserves investigation before it becomes a success metric.

This exercise often exposes duplicate signals. A purchase may arrive through two tracking sources. A form event may fire on a button click even when validation prevents submission. Verify the event itself rather than relying only on its name.

Check what bidding is learning from

Google distinguishes primary conversion actions from secondary actions. Primary actions can be used for bidding when their standard goal is selected; secondary actions are generally observational. Custom goals are an important exception: included secondary actions can also be used for bidding. Inspect the campaign’s effective goals, not just the action’s label. Google’s conversion-action guide explains this distinction.

Changing those settings is a measurement and optimisation decision. Capture the previous configuration, understand the affected campaigns and agree how the transition will be evaluated.

Keep the before and after comparable

Imagine a fictional account previously counted both enquiries and brochure downloads. After an audit, only valid enquiries remain in its main outcome measure. Reported conversions fall. That does not, by itself, mean advertising deteriorated: the definition changed.

Mark the date of the change and establish a baseline under the new definition. Keep the old and new measurement periods separate when comparing cost per acquisition. Check reporting delay before treating recent outcomes as final.

Your next useful move

Choose one campaign and reconcile its reported outcomes with the underlying business records. Document what matches, what does not and what remains unknown. The deliverable is a trusted outcome map and a bounded list of gaps—not a promise that a tracking repair will create more customers.

Atlas reading: S01 — Trustworthy measurement; Data and Evidence Contract; MEAS-01 — Audit conversion actions.